N6-methyladenine (6mA) is a modified form of the DNA base adenine, where a methyl group is attached to the sixth position nitrogen base. The genomes of several kingdoms, including bacteria, fungi, plants, and mammals, contain this epigenetic alteration. Contrary to the well-researched 5-methylcytosine (5mC), the function of 6mA in molecular biology is yet to be explored. Recent studies suggest that 6mA may impact the integrity of the genome, DNA repair, and transcriptional control. Understanding the functions and mechanisms of 6mA can provide valuable insights into epigenetic regulation and its implications in development, disease development, and evolution.
MethSemble-6mA is a 6mA prediction tool developed by combining ensemble machine learning, hybrid feature selection and bootstrap samples. The tool utilizes five different feature sets for DNA sequence vectorization: di-nucleotide frequency, GC content, AMIP, mono-nucleotide binary encoding and nucleotide chemical properties. Nine machine learning models, including support vector machine, random forest, k-nearest neighbor, artificial neural network, multiple logistic regression, decision tree, naïve Bayes, AdaBoost, and gradient boosting, were employed with relevant features selected through the feature selection module. The top three best-performing models, gradient boosting, random forest, and SVM, were then combined into a robust ensemble model for predicting sequences with 6mA sites.
Copyright © 2023 ICAR-Indian Agricultural Statistics Research Institute, New Delhi-110012. All rights reserved.